A changed welfare-state: Sweden through the 1980s

Transcription

1 A changed welfare-state: Sweden through the 1980s How the organization of labor unions and politics matter for economic performance Ole Henrik Hunstad Vik Department of Comparative Politics University of Bergen Spring 2015

2 Abstract This thesis seeks to contribute to the important debate regarding whether who governs matter, and if the relationship between the government and labor unions affect economic performance. The research question is: "Did changes in the organization of labor in the 1980s lead to changes in the partisan influence on economic outcomes in Sweden in the 1990s?" To answer this a case-study of Sweden centered around the 1980s is conducted, a period in which this highly social democratic country underwent several changes to its economic system; a shift away from its corporatist roots. Corporatism, the cooperation between government and labor unions, is first examined theoretically through identification of partisan strategies for growth and class alignment. These are situated in a European context between 1975 and 1995, and the Swedish welfare-model is discussed. Further, the case-study shows how Sweden developed from a post-war welfare-system to a notably less corporatist system by the end of this period. Sweden is historically regarded as an example of a markedly corporatist country, and in Europe it is thus an unexpected case of a decline in corporatism. The thesis shows how changes to the organization of labor in the 1980s had consequences for economic performance, and this performance is discussed in the context of Europe, Scandinavia, and Sweden. The economic performance of the 1990s is related to changes in the 1980s, and the development between the periods analysed using a path-dependent methodological perspective. The development illustrates how the decline in Swedish corporatism had implications for subsequent performance. In the 1980s this is best explained by the dissolvement of centralized bargaining between labor unions and employer organizations. The institutional linkage between the leftparty and labor, which remained strong from the 1950s to the start of the 1980s, appears weakened as a result of shifts in government power. With this came new policies, and I

3 the economic system moved towards market liberalism. Economic turmoil increased in the 1990s, starting with a banking crisis spanning Scandinavia, and the cause of this is traced to policies of the preceding decade. In a country with a long history of stable economic performance through cooperation between a leftist government and an encompassing labor, these changes and shocks were not expected, but they point to a broader decline in corporatism; a result of economic crises and changes to the political control over the economy. II

4 Acknowledgements After a year of writing I owe thanks to several people who helped in making this thesis a reality. For her patience and thoughtful discussion on the topics contained within these pages, and for valuable feedback on many drafts I thank my supervisor Associate Professor Elisabeth Ivarsflaten. Advice given by Associate Professor Jan Oskar Engene whilst supervising my initial research proposal were also very helpful. The academic staff at the department for Comparative Politics and their efforts have been greatly appreciated in my time here, and I have enjoyed attending lectures, seminars, and meetings, as well as holding seminars myself. My fellow students at Sofie Lindstrøms hus, without whom this thesis would likely not have been finished, deserve thanks for always being in high spirits. Long days and nights were spent writing and talking, all of which were enjoyable. Finally, my family deserve thanks for their patience and support, and also my wider social circle for providing much needed breaks from work on the thesis. My disposition towards the progress of writing it was not always sunny, but you kept me motivated. III

5 Contents Abstract I Acknowledgements III Contents IV List of Tables VI List of Figures VI 1 Introduction Academic relevance Corporatism and contribution Structure of the thesis Theory Theoretical scope and perspectives Democracy and capitalism Who Governs: The economic impact of political latitude The virtue of collective bargaining Achievable economic performance Class and political alignment Literature Review Wage-restraint The Garrett et al and Jackman debate Method Studying corporatism Sources of data Analytical framework Economic performance Left-labor linkage Methodological Approach Validity Changes in labor Encompassing labor IV

6 Contents 4.2 Centralized bargaining Rehn-Meidner model of wage-bargaining Effects of centralized bargaining Towards decentralization The effect of a decline in corporatism Viability of the argument An updated view of the explanatory model Economic performance Expectations Decline in corporatism Results Growth Unemployment Inflation Implications for analysis Analysis Left-labor linkage and overlap Performance Path-dependent changes Policies leading to the crisis The crisis Implications and consequences Causality Corporatism Conclusion 76 Bibliography 83 V

7 List of Tables 1 Political strength of the left, labor unions and economic growth List of Figures 1 The relationship between government cohesion, encompassing labor, and economic performance Union Density and Coverage Government Cohesion in Sweden An updated relationship between government cohesion, encompassing labor, and economic performance Growth rate in the EU, the OECD, Scandinavia, and Sweden Unemployment in the EU, the OECD, Scandinavia, and Sweden Unemployment in Scandinavia Union Density and Coverage in Scandinavia Inflation in the EU, the OECD, Scandinavia, and Sweden Left-labor Linkage Performance in Sweden Wages and inflation in Sweden A revised relationship between government cohesion, encompassing labor, and economic performance VI

8 1 Introduction The question "Does Who Governs Matter?" have been asked in many forms over the years by students of politics (see Lange and Garrett 1985; Fox 1988; Cameron 1988; Esping-Andersen 1990; Drezner 2001). In this thesis I aim to answer this question by studying the relationship between government and labor. My research question is the following: Did changes in the organization of labor in the 1980s lead to changes in the partisan influence on economic outcomes in Sweden in the 1980s and 1990s? I argue that the relationship between the government and labor in Sweden changed in the 1980s. The central argument of the thesis is that economic outcomes are explained by changes in labor unions, not just partisan policies, and that a change in the mechanisms of this relationship also changed political latitude to shape economic policy. Parties of the left and right necessarily have to maneuver this political space when formulating policy, and changes in these mechanisms explain the political impact on the economy. I will examine whether developments of the 1980s in the economic system and the organization of labor have changed economic performance in a country typically regarded as corporatist in nature. The thesis relates to present-day debates on the state s role in the economy and how economic policies impact welfare and growth. To identify both the impact and the changes the problem must be contextualized properly: I examine the period prior and subsequent to the 1980s to show the development of corporatism historically. Sweden is selected because of its prominent position as highly corporatist, both in comparison to the other Nordic countries and Europe (Esping-Andersen 1990: 114). 1

9 The development between the periods take the form of a historical narrative, where changes are analysed as path-dependent that is, to some degree inevitable. For clarity I distinguish systemic changes from external shocks to the economy, as Sweden experienced both. Though the approach is mainly qualitative in nature, quantitative data is also employed to illustrate the impact of these changes and shocks. The data ranges from the 1950s to the end of the 1990s where available to fully show these impacts in a broad historical context. This is then a longitudinal case-study of Sweden centered around the 1980s and early 1990s, a period in which structural political and economic changes occurred, as well as economic shocks. Institutional impact is examined and analysed alongside the political and economic, to account for the relationship between the government and labor. The ultimate goal is to show how the Swedish welfare-state developed, and how the changes and shocks of the 1980s and 1990s impacted this. 1.1 Academic relevance My initial focus when I started working on this thesis was the long-term development of economic inequality and its relation to policy, a broad topic which I approached with a broad mixture of qualitative and quantitative methods, on multiple empirical levels. When it became clear that covering this from the 1980s and into the 2000s could not sufficiently be done in pages, I moved towards studying economic policies of the 1980s. After an evaluation of countries which are clearly relevant in this vein the United States and England it became necessary to select a case, or limited number of cases, in which this development was less obvious. In terms of policies, historical position, and empirical material, this would be the Nordic countries. As explained in the case-study and the analysis, Sweden was the most prominent case of social democratic corporatism in the examined time period, and accounting for changes within this political system helps to identify structural changes which reveal a potential decline in corporatism. 2

10 1.1.1 Corporatism and contribution Corporatism is often associated with the bureaucratic functions within institutions, and how they relate to the state. I use it as a concept of interaction between labor, capital, and the state, thus explaining the role played by interest-groups and the state in determining economic policy. It is applicable to a broader analysis of the relationship between the political and the economic, and how this relationship can constrain the political latitude for economic policy. For many the 1980s is a historical period of economic deregulation as a response to the laggard growth of the 1970s. Contextually, the most apparent examples of this was under Ronald Reagan in the United States and Margaret Thatcher in the United Kingdom. My interest in Sweden is motivated by what it appears to have been in this context: A case of a country least likely to follow the policies instituted by the aforementioned countries, and one which for most of the post-war period had sustained an economically viable welfare state. As will be shown, the cooperation between the left side of politics and the labor unions helped sustain both stable growth and low unemployment. The contribution made to the field of comparative political economy is first of all a retrospective on the development of Sweden as a welfare-state and in terms of economic performance. As it was a country least likely to follow the development of the more economically liberal countries, the changes identified are all the more important. This development was unexpected in a historical perspective, as it is along with its Nordic neighbours regarded as an extensive welfare state. On the surface, the economic performance of Sweden throughout the 1970s does not indicate a need for systemic change to its economy. Further, the relevance for the "real world" (see King et al. 1994: 15) of this thesis is that the empirical and systemic situation of Sweden prior to the 1980s appears as an alternative to the current status quo of limited political influence on the economy. The focus on low unemployment in combination with stable inflation and high growth, as was the model from the 1950s to the late 1970s, should be desirable for many countries, and importantly this performance was sustained for a long period of time. Finally, I find that Sweden as a case of a decline in corporatism is a perspective on economic inequality. The cooperation between the left and labor for a long time ensured 3

11 that growth in wages followed increases in costs of living. In the long-term this became unviable, as I discuss throughout the thesis, but the larger question is whether labor unions could again function as a check on wage-differentiation. That is, the development of wages which occurred in Sweden prior to the 1980s may be a desirable way to lower income inequality also in modernity. 1.2 Structure of the thesis In chapter 2 relevant theoretical perspectives are discussed, and necessary assumptions and conditions for the empirical relevance of an explanatory, general welfare-model specified. These are then related to the research question, and a literature review is conducted. Chapter 3 outlines how corporatism is studied, what sources of data are used, and the method applied. An analytical framework is specified, and the validity of the chosen method discussed. The first half of the case-study is in chapter 4, where the relationship between the government and labor is further developed. This includes a historical narrative and a discussion of whether a decline in corporatism can be identified. The second half of the case-study is in chapter 5. Here Sweden is placed in the context of Europe and of economically advanced democracies, and subsequently Scandinavia, to show how economic performance changed over time. Based on this implications for the analysis are discussed. In chapter 6 the development is examined in terms of path-dependence and causality. The implications of this, and the consequences for the Swedish case and welfare-model, are discussed and a decline in corporatism considered. 4

12 2 Theory The left-right divide in economic policies is typically identified by diverging strategies for achieving economic growth: The right follows a conservative strategy of tax-cuts, and the left a liberal strategy of increasing the production rate of labor and capital (Boix 1998). A facet of this is the degree to which government can and will intervene in the economy, where the right is expected to pursue a laissez-faire doctrine whilst the left is expected to intervene to ensure more redistribution of wealth and risk (Alvarez et al. 1991: 539). Thus, from the outset partisan policies are assumed. A trade-off supposedly exists where less intervention yields more efficiency and more intervention yields more security and stability. The implication is that rightist government prefers freer markets whereas leftist governments prefer more social security. Parties preferences for market freedom are constrained by the viability of their political strategies that is, policies for achieving growth. That is to say, who governs matter for economic performance mainly where there are differences in economic policy, and these policies are limited by what policy can achieve in a relatively free market. To identify the development of these constraints over time, I evaluate the role played the government in the economic changes of the 1980s, expressed as a model of political latitude to instigate economic policy. The development of this model is later applied to distinct periods of historical development, where the application of the model relies on a set of assumptions and conditions, as defined below. This chapter is structured as follows: I begin with an outline of the theoretical scope of the argument, the assumptions it relies on, and how the theoretical perspectives relate to the research question. Further, a literature review is conducted which focuses on two prominent theoretical perspectives, their implications for the argument, and specifically on a debate between Garrett et al and Jackman. 5

13 2.1 Theoretical scope and perspectives Two concepts should be clarified immediately: Corporatism and latitude. The latter is defined as the scope for freedom of action, and in this thesis refers to the possible policychoices available to politicians at any given time. It is the policy-choices politicians can make which directly affect the economy, essentially the political control over it. Thus it explains the leeway or maneuverability politicians have in relieving economic problems through political means. This concept also includes a negative scope in the form of constraints on political action. Note that a line will not be drawn between latitude and constraints, as the concept does not lend itself to such clarity. Rather, it is an evolving perspective on what can and can not be achieved with economic policy, and thus highlight how and why politics matter for economic performance. An example of political latitude would be the governments ability to change the inflationrate by setting an interest-rate for the central bank to use. Policy changing this rate to deal with inflation or unemployment, as discussed below are an important tool in controlling the stability of the economy, especially against shocks. Politicians ability to change this rate would be constrained by an independent central bank, as discussed in section Hence, in this thesis political latitude describes what policy can and can not achieve in regards to controlling the economy. Corporatism is often associated with political systems or bureaucratic functions, but its use here follows Schmitter (1974) and is the institutional linkage between government and labor. More specifically, this thesis evaluates it as a dimension with social democratic corporatism at one end of the spectrum and market liberalism at the other. It is the interaction between a government of the left or right with labor, as well as the governments preference for such a linkage, and what policies this entails. As the difference in economic performance is judged empirically, the use of the concept avoids a normative interpretation. Corporatism is then defined as a political system in which interest-groups play a part in policy-making, and through this express the preferences of their members for economic policies. Further, economic performance is a composite empirical measure consisting of growth, unemployment, and inflation. It will be referred to as a general phenomena at various points throughout the thesis, but it is assumed to always be sensitive to systemic changes 6

14 and economic shocks, which are specified empirically. The interaction between government and interest-groups, expressed empirically through wage-bargaining with labor unions, defines the characteristic mechanism of social democratic corporatist systems. The aforementioned model defines the setting in which the central argument applies. First, the setting for the analysis is a political situation with free elections and a relatively free economy that is, a capitalist democracy (see Schmidt 1982a); Second, parties of the left and right have partisan policies which they will pursue if elected, and as such primarily seek election and secondarily seek to implement these policies if elected; Third, elected governments have the political capital and capability to implement these policies, and can achieve the formulated goals of their policies with some success; Fourth, the political and economic constraints may vary a lot when applied to specific countries, but can be deduced from theoretical perspectives and verified by case-analysis; Fifth, the policy-preferences of the left and right reflect the policy-preferences of voter-groups, and will in highly corporatist systems also be reflected through interest-groups which political parties align with. In what follows, each of these conditions are developed and their relation to the central argument specified Democracy and capitalism The analysis is limited to democracies of a capitalist nature, and the reason for this is two-fold: Examining differences in partisan policy necessarily requires parties with distinct policies. A prerequisite for distinct policies is first of all a competitive electoral regime, where electoral victory gives control of government. Second, if the policies of the competing parties are similar or near-identical, there is little value in voting for any of them as the outcome would not differ. Hence, the democracy must in Dahl s terms allow both public contestation and all citizens to participate in elections and office (1971). The other facet of this is capitalism, which is presumed to be necessary because its essential function is to facilitate growth (Schumpeter 1942). As the dependent variable is economic performance, of which growth is a part of, variance in this outcome must be present for an analysis to be valuable. This does not, however, exclude countries whose 7

15 economies do not perfectly resemble the capitalist ideals of the United States. To account for why who governs matters a comparative ideal must be established, and as my focus is on the impact on economic performance, a high degree of democracy and a high degree of capitalism must be assumed. There is an inherent question of what high degree implies here, as neither democracy nor capitalism are necessarily binary concepts (Collier and Levitsky 1997). I will expand upon this point in section 2.2, but for now it should suffice to say that a high degree of both will produce a situation with political competition and relatively variable economic performance, thus excluding political systems of varieties such as one-party systems, authoritative systems, plan-economies, or the like Who Governs: The economic impact of political latitude As specified above, political parties are assumed to have distinct economic policies. This assumption is necessary if there is to be any applicability to the argument, but more importantly if there is to be any value to politics. Even if who governs matters, what explains partisan efficacy in attaining economic performance must be answered. If the economic factors growth, inflation, and unemployment are assumed to vary conjuncturally as a structural aspect of a capitalist economy, what difference the economic policy of the left or the right makes is left unanswered. First of all, theory dictates that elections are not won just on the basis of policy-ideas (Jackman 1987: 255). This is to say that parties must not only convince voters with their ideas for policies to implement, but likely also make alliances between parties and with voter-groups. An electoral victory does not itself guarantee that policies are implemented, hence government strength must be considered explicitly. Passing any legislation requires a sufficient majority of seats in the legislature, and thus government cohesion as defined in section 2.2 explains economic performance by considering the political power necessary to pass legislation. Furthermore, it may be necessary for parties to cater to voter-groups whose political preferences align with their own. These voters are often represented by, or members of, an interest-group who will lobby or otherwise try to influence the politics of the governing party or align themselves with some other party. Crucially, and as fronted by 8

16 Garrett et al (Lange and Garrett 1985), the alliance between the left and encompassing labor explains economic performance aside from economic factors. I return to their analysis in section The virtue of collective bargaining Interest-groups seek to increase their utility by following one of two strategies, as specified by Olson: "[...] the organization can in principle serve its members either by making the pie the society produces larger, [...] or alternatively by obtaining larger shares or slices of the social pie for its members" (1982: 42). However, note a caveat as far as incentives are concerned: "[...] the incentive for group action dimishes as group size increases [...]" (Olson 1982: 31, original emphasis). If we consider how large labor unions can distrupt the economy through wage-militancy, it is easy to see that the self-interest of each organization can affect economic performance negatively. The exception to this, as specified by Olson (1982: 48) and as extended by Lange and Garret (1985: 795), is when labor is encompassing and thus has a bigger impact on the economy, as explained below. The economic cost of policy for the strategy preferred by either side of politics must be considered, because welfare measured for example by unemployment and inflation is a trade-off between social security and economic freedom, and trying to increase either then implies a cost to the other. If parties do not vary in their preference for one over the other, a well-meaning dictator may achieve the same levels of both. This is to say that for politics to matter for economic performance who governs must necessarily matter, and because there are tangible differences in empirical results for both welfare and growth, partisan politics must be counted as an explanatory factor. The implication of this is two-fold: First, any constraints on policy-making will affect both sides of politics, as they limit what can be achieved in both welfare and market freedom. Second, an analysis of empirical outcomes is contingent upon all of the aforementioned assumptions holding, as well as a comparative focus. How much who governs matter is likely to only be clear if governments of the left and right are seen in relation to one another, which necessitates an empirical evaluation. 9

17 2.1.4 Achievable economic performance If low unemployment is seen as a good for welfare and low inflation as conducive to stable economic growth, then a trade-off between the two imply a contestable area for policy. This trade-off is known as the Phillips Curve (Phillips 1958), and it has a corollary in politics in the sense that the trade-off is affected by political decisions (Hibbs 1977). Assuming that parties on the left are unemployment-sensitive and parties on the right are inflation-sensitive (Alvarez et al. 1991: 540), the political economic Phillips curve explains the policy-choice where lowering unemployment increases inflation, and vice versa. In addition to being a constraint on economic policy, this trade-off is a mechanism by which parties can counter economic shocks in the short-term. Note that a common distinction is to say that in the short-term few variables change, whilst in the mediumterm several variables can change, and in the long-term everything can change. It is also important to note that in the long-term the viability of this trade-off decreases, as explained below. Central bank influence is not explicitly treated here, but whether government directly controls inflation or cedes the responsibility of this to a central bank does affect this trade-off. Since central bank independence is largely a phenomenon of the 1990s and beyond (Glyn 2006: 32), the occurrence of it in a potential case-study must be considered. The vertical Phillips-curve An argument has been made for a viable strategy for full employment (Kalecki 1943; Boddy and Crotty 1975), based on ideal conditions and with a domestic focus. Empirically it is evident that no advanced democracy has thus far produced full employment (International Monetary Fund 2014), in the sense of no unemployment. More importantly, full employment means that everyone who can be employed are. These democracies have had their share of leftist governments, many of which have worked towards reducing unemployment, but none of which have abolished it. An explanation for this is that Marx prediction of a Reserve Army of Labour still holds. This reserve army is a by-product of capitalism, in that its existence is explained by the 10

18 willingness of the unemployed to work for less wages than the currently employed, doing the same work (Stilwell 2006: 124). Surplus, or increases in, population thus sustain and renew this reserve army, and as long as the supply of labor exceeds the demand for labor, it is a mechanism that keeps "[...] wages down and profits up" (Stilwell 2006: 124), and it is inherent to capitalism (Marx and McLellan 2008: ). I will not delve into Marx critique of capitalism, but rather note that this mechanism is of great importance as it highlights a facet of the unemployment-inflation trade-off. If unemployment is considered a disadvantageous situation for those who currently are, but also as a loss in efficiency due to the superfluous labor available, it should clearly be in the interest of either side of politics to keep unemployment low as it is beneficial for the output of the economy. However, because this mechanism is inherent to capitalism it can be considered a natural unemployment, that is, structural unemployment. Long-term unemployment is often explained by a long-term Phillips curve, where efforts to lower unemployment consequently will lead to a state where inflation is increased but unemployment returns to the previous level, thus rendering full employment impossible without spiralling inflation. This, however, is in the indefinite long-term, and reducing unemployment is still a viable policy-choice in the short- to medium-term Class and political alignment Below I discuss the empirical relationship which suggests that centralized, well-organized, and politically strong labor movements in combination with a coherent leftist government will yield better economic performance than if the labor movement was weak. Why this should necessarily be the case is uncertain, as interests-groups are not necessarily in the business of increasing a society s output. Interests-groups, whether they are small special-interest lobby groups, worker unions, or employer-associations, have one purpose: To protect the interests of their members by ensuring them increased utility relative to the cost of membership. Economically, this is fairly self-explanatory. If the cost of belonging to a union outweighed the benefit, it would be irrational to remain a member. This follows from Olson s strategies for interest-groups. 11

19 Further, labor unions are mainly interested in ensuring a living wage and increased welfare for their members. Apart from worker rights, which are not covered here, they do this by wage-bargaining with employer-associations, and this affects the economic output of society. Since labor unions are responsible for the most part only to their members, there is no reason for why they should restrain themselves in this wage-bargaining, regardless of potential costs imposed on society. The implication is the following: If the interest-group manages to redistribute more resources and means to its own members, then it does so at the cost of the rest of society. Consider, for example, a small labor union which through wage-bargaining secures an increase in wages. The cost of the wages paid to the workers which the labor union represents will either be borne by the owner of the business, or passed on to society in the form of an increase in the price of the goods which the business supplies. All things equal, it is assumed that increased wages will yield increased prices, and as such wage-bargaining affects the aggregate price-level in the society. That is, unless the size of the labor union is encompassing (Olson 1982: 44). Given a sufficient size of the labor union, whether through a large single organization or a coalition of unions, and a rational leadership, the union will also be interested in increasing the productivity of society at large. Given the size, which will vary between societies, the aforementioned cost to society would then be far more likely to also be a cost to its own members since it encompasses a much larger part of the society (Olson 1982: 48). Hence, to avoid negative consequences for its members, it has an incentive to show restraint in wage-bargaining and rather focus on making the pie society produces larger (Lange and Garrett 1985: 795). 2.2 Literature Review In what follows I examine how wage-policies, as a mechanism of cooperation between the left and labor, help explain economic performance and then examine how the work of Garrett et al and Jackman aid in explaining partisan differences in this outcome. 12

20 2.2.1 Wage-restraint From the outset there are no obvious reasons for specific outcomes of a wage-bargaining process, as it is highly dependent on the contextual nature of the current wage-level as it relates to overall economic performance. To find an explanation for these outcomes, the cooperation between political parties and interest-groups are often examined in corporatist systems, as this is an area where wage-bargaining yield wage-policies. A methodological issue with such an examination is that wage-negotiation, whether it is between labor and government or labor and employers, is inherently difficult to measure: The outcome may be relatively easy to ascertain, as wages change relative to the price-level, but the process itself cannot be quantified to yield the exact outcome. In an effort to explain what determines wages, be it purely economic factors such as growth, unemployment, or inflation, or whether wage-negotiation and thus politics play a part, Armingeon proposes the following hypothesis: [...] there will be political wage restraint if union bargainers are influenced by the social democratic party and if the rank and file fail to protest against, or even agree with, this policy (Armingeon 1982: 235). In this regard wage-bargaining happens through an institutional linkage between the government and labor, where particularly the cooperation between labor and the left manifests. Three major types of this linkage can be identified (Armingeon 1982: 235): (1) Collective affiliation: Union members automatically become party members. (2) Ex officio representation: By the virtue of position or status union leadership has a place in the party executive, or vice versa. (3) Liason committees: Co-ordinated committees involving both the labor union and the party. Empirically the first clearly has some merit, as it existed in Norway, Sweden, and the United Kingdom, making it an important area for case-analysis, whilst the second and third more resemble channels of information exchange (Armingeon 1982: ). If collective affiliation can be determined to be present as a mechanism or in high levels in the empirical period in question, then any of these countries would be suitable for determining partisan impacts on economic performance. I return to the left-labor linkage 13

21 and involvement in the wage-setting process in section The Garrett et al and Jackman debate Between 1985 and 1991, a debate ran between a handful of authors on the efficacy of economic policies in industrial democracies, and how partisanship, government cohesion, and labor unions affected economic performance between 1960 and 1984 (Lange and Garrett 1985; Jackman 1987; Lange and Garrett 1987; Jackman 1989; Garrett and Lange 1989; Alvarez et al. 1991). Garrett et al made the argument that governments of the left and right have distinct economic policies which they want to implement, and whilst the left is more focused on keeping unemployment low, the right is focused on keeping inflation in check. This is consistent with the assumption that either side of politics have different strategies for achieving economic performance. Their model explains how a political alliance between parties of the left and organized labor can yield economic performance on par with what would be expected from a free market economy. Additionally, they find that the presence or absence of organized labor explains strong and weak growth (Lange and Garrett 1985). My aim in this section is to broadly describe the theoretical and empirical basis of their argument, and based on this develop a comprehension of why these configurations potentially limit political latitude. Labor as a social mechanism Following Olson s logic of collective action by interest-groups, the authors focus on the propensity of the labor unions to self-regulate that is, show wage-restraint. As shown previously, such behavior is rational for large unions given that they are large enough to be encompassing. As Lange and Garrett point out, this crucial size is uncertain because the social- and group-utility of showing such restraint is not determined by the unions themselves, but rather by the owners of the businesses in which the union-members work (Lange and Garrett 1985: ), and it will obviously vary between countries. In contrast to the laissez-faire approach of the right, the left can positively affect both the wage-negotiation process and the uncertainty of the unions. In this regard who governs 14

22 clearly matters, as it serves not only as a potential mediator of these negotiations, but also as a distributor of economic growth and thus a guarantor of increased welfare without necessarily increased inequality. There is a dual contingency to this mechanism: Enhanced economic performance is only likely when strong labor unions support a strong leftist party, and in instances of a weak labor and a strong rightist party. As the welfare-model strongly relies on wage-bargaining taking form between an organized labor union and employers, mediated by government, this precondition for enhanced performance is essential to evaluating the model. Lange and Garrett s summary is reproduced in table 1, where they also outline four related hypotheses marked A to D. If economic performance is not stronger under scenario A and D than B and C, then the predictive and explanatory power of the model will be low. Table 1: Political strength of the left, labor unions and economic growth Political Strength of the Left Encompassing Organization of union movement Not encompassing +: positive contribution to economic growth -: negative contribution to economic growth Source: Lange and Garrett 1985: 799. Strong (A) + (C) - Weak (B) - (D) + The four hypotheses can briefly be explained as the following: A When the left controls the government, they reduce uncertainty and hence the risk of wage-restraint, to the degree that more distribution occurs as unions focus on making the societal pie larger, rather than obtaining a larger slice of it (Lange and Garrett 1985: 799). B With the absence of an assuring left party, unions focus more on obtaining a larger slice of the pie for their members. Thus wage-restraint is not a beneficial strategy 15

23 in the short-term of an election cycle, and there are no gains from hoping that business-owners or a government of the right would distribute economic growth as they are by definition less likely to do so in comparison to a government of the left (Lange and Garrett 1985: 800). C In a situation with no large union, or coalition of unions, each of them will have little to gain from pursuing any other strategy than to increase the utility of their members. This is largely because unions bargain with their respective employer, not on behalf of a larger labor market, and thus have a lesser impact on wages in aggregate (Lange and Garrett 1985: 800). D When a strong right government has no organized or significant opposition from labor unions they can shape the economy closer to the ideal of a free market, thus creating economic performance at the cost of social distribution (Lange and Garrett 1985: ). These are relevant for the empirical analysis, but it should already be noted that as Lange and Garrett put it a linear relationship between economic growth and the presence or absence of labor power should not be expected (1985: 801). This is to say that labor is likely not the only explanatory factor for comparatively high economic performance, and there might be cases where growth is achieved in spite of not having a particularly strong labor movement or strong government. Performance as the dependent variable Initially, Garrett et al used economic growth as the only indicator of economic performance (Lange and Garrett 1985: 802), and it is also the dependent variable in their analysis. Later, this performance included "annual rates of gross domestic product growth, change in the annual rate of the consumer price index growth, and change in annual employment levels" (Alvarez et al. 1991: 545). The period they examined was 1960 to 1980, and changes in performance within this time-frame. Lange and Garrett sought to identify the impact of "[...] labor on politicaleconomic performance outcomes from [...]" (1985: 802) in light of the growth between 1960 to There are methodological issues with this, as addressed below, 16

24 but it should immediately be noted that the oil shock of 1973 highlights a potential confounder to the variable, namely external shocks. Economic shocks The degree to which a country depends on imported goods and gain much of their wealth from exports determine how they are affected by an economic shock. Garrett et al (1991) account for two varieties of changes in economic realities: Shocks and systematic changes to economic processes, both of which I apply throughout the thesis. The former is named Vulnerability to International Economic Conditions and controls for conditions external to the economy, whilst the latter is named Domestic Economic Processes, which are changes in economic performance lagged to show changes in the domestic economy. These variables reveal not only whether economic performance was affected by factors other than the composition of government, but also whether government changed the structure of the economy. And, all else equal, they can verify that even under extraordinary empirical circumstances such as a shock in oil prices their argument holds. In the analysis I discuss this as changes and shocks to the economic system. Government Cohesion In Garrett s words "[...] economic policy should be more consistently partisan and macroeconomic performance should be stronger the more coherent conditions are in the politics and the labor market" (Garrett 1998: 26). This is to say that efficient policies are contingent on strong parties, where this strength is measured by a Government Partisanship variable, or how many seats of the parliament cabinet the governing party holds (Alvarez et al. 1991: 545). The more seats held, the more political power the government has, and hence the easier it would be to implement policies. Encompassing labor Alvarez, Garrett, and Lange use the variable Labor Organization which they measure by the density and centralization of union movement (1991: 545). That is, whether the country s labor movement is sufficiently large to impact wagenegotiations, and hence the economy at large. Additionally, it is aware of this impact and will voluntarily restrain itself and prefer long-term wage growth over short-term, as this growth impacts economic stability. This is conducive to policy-makers as militant 17

25 wage-negotiations could by itself create inflation. This variable also includes the opposite of an encompassing labor: Weak labor. In this case, labor is "[...] sufficiently weak and decentralized to be rendered unable effectively to retaliate against the imposition of market discipline [...]" (Alvarez et al. 1991: 543). Here, governments of the right will have less resistance when imposing market liberalism. The strength of this relationship between political parties and labor is contingent on the strength of the party itself. The Jackman critique The argument forwarded by Lange and Garrett was critiqued by Jackman (1987) for its empirical content and assumptions, as well as for the theoretical linkage between parties and policy. Their model is based on economic growth between 1974 and 1980, but this is calculated based on the relative growth between 1960 and Though not necessarily an invalid approach, Jackman raises the question of whether the earlier period should be an explanatory factor on line with labor strength and government cohesion. Whether thirteen years of growth is comparable to six years, or if a weighted average would be a more precise measure than simply average annual growth in each period may be the case (Lange and Garrett 1985: 802). Their sample of countries is also limited as it includes sixteen "advanced industrial democracies" (Lange and Garrett 1985: 794), and a bias is introduced by including the case of Norway. Finally, the effects of leftist strength is questioned, as their results run contrary to similar empirical analysis. If government cohesion was insignificant, the question of "does who governs matter?" would be reduced to "does labor matter?". However, Jackman does concede their central argument, which is: Through its policy instruments, government can influence both the investment behavior of capital and the distribution of economic growth among social classes (Lange and Garrett 1985: 797). As developed in section 3.3.3, my approach applies their argument using an inherently different methodology, wherein these issues of quantitative analysis are resolved. Jackman further draws the link between their argument and Lipset s (1960) view that elec- 18

26 tions express a class struggle. Whilst his critique of the assumption that the "[...] state is directly in the hands of labor [...]" (Jackman 1987: 254) is justified, a more nuanced explanation is possible: The state is not necessarily in the hands of anyone, but the preference-alignment between the left and labor may still explain a prevalence in policies which are favored by both. The alliance between the left and labor does not imply that either is dominated by the other, but rather a symbiotic relationship in which labor fulfils its responsibilities to its members whilst the left gets elected. The validity of the assumption that partisan politics will yield distinctive policies are challenged by Downsian policy-convergence, where parties propensity to be primarily office-seeking and secondarily policy-formulating lead to a converge on the preferences of the median voter (Downs 1957). Thus, policies would grow ever closer and less distinctive in order to capture the swing-voters, who reside around the median voter. Without delving into a technical explanation of the median voter theorem, suffice it to say that it is a paradox resolved by the empirical variance in policies and by identical policies disincentivizing voting by leaving no variance in the outcome of an election. Even if both sides of politics aim for economic growth, the path taken to achieve this can and will vary. Even without any differences in the outcome for growth itself, social factors such as distribution can be different without necessarily impacting growth negatively. Hence policy-convergence of the Downsian sort is unlikely, as both political theory and the model assumes a distinctiveness to policies of the left and right. Causality, conditionality, and mechanisms In Garrett et al s response (1987) to Jackman three points are worth reiterating and evaluating: An argument of causality, an argument of conditionality, and what either means for the central argument. Though the combination of a strong left government and a strong labor union leading to economic growth may be a positive causal relationship, it is important to recognize that this hinges on identifiable mechanisms to show this, not on most of the empirical materials either side of this debate presented. As sophisticated as their methodological approach is, it does not lend itself to a causal argument, because their quantitative data does not account for the underlying mechanisms of the issue. The mechanisms of this relationship will be explored further below, along with an evaluation of whether a causal relationship holds. 19

27 Their conditionality-argument, however, does seem plausible, and further it also explains the case of Norway not as a deviant case, but as a potentially crucial case (see Gerring 2007). As already explained, it is not simply the co-existence of a strong left government and strong labor, but rather an interaction between them that would lead to relatively strong economic performance. As Garrett et al point out, "Norway s exceptional growth in fact reflects its position as both a highly corporatist and oil-rich country" (Lange and Garrett 1987: 260, original emphasis). Hence, Norway exemplifies how the strong ties between the left and labor manifested in strong economic performance, as they became oil-rich through government efforts (Jackman 1987: 245). In a restatement of why who governs matter, Garrett et al suggest the following: The argument is not that political conditions totally determine economic outcomes, but rather a more modest assertion that politics affects economics at the margins, and that these margins are nevertheless very important (Lange and Garrett 1987: 269). This emphasizes an important point, namely that whilst who governs matters, it is not the only explanation available. Technological and societal factors, as well as economic shocks, will all impact the economic performance of various countries. The assumptions for the explanatory model implicitly involves some degree of dependency on foreign markets, and this alone will ensure fluctuations in the domestic economy. Beyond the original argument After the initial debate, revolving mainly around methodological and empirical issues, Hicks (1988) respecified the model proposed by Lange and Garrett (1985; 1987) and still found that the empirical data supported their argument. In what follows, I will outline this respecification, evaluate its merits, and discuss responses by Jackman (1989) and Garrett et al (1989; 1991) as well as the further implications. Hicks contributions First, Hicks expanded the model to include two alternative interpretations: Either labor organization effects are conditional on labor s political strength, or labor s political strength is conditional on union strength. But this does not suggest that either may produce a "[...] distinct causal process" (Hicks 1988: 682, 20